America Projected to Triple Semiconductor Manufacturing Capacity by 2032, the Largest Rate of Growth in the World

[Daniel Nenni]

U.S. forecast to increase its domestic chip manufacturing by 203% in the decade following enactment of the CHIPS and Science Act—and expand its share of the world’s total fab capacity for the first time in decades—according to new Semiconductor Industry Association/Boston Consulting Group report focused on the global chip supply chain

WASHINGTON—May 8, 2024—The Semiconductor Industry Association (SIA), in partnership with the Boston Consulting Group (BCG), today released a report on the global chip supply chain that projects the United States will triple its domestic semiconductor manufacturing capacity from 2022—when the CHIPS and Science Act (CHIPS) was enacted—to 2032. The projected 203% growth is the largest projected percent increase in the world over that time.

The study, titled “Emerging Resilience in the Semiconductor Supply Chain,” also projects the U.S. will grow its share of advanced logic (below 10nm) manufacturing to 28% of global capacity by 2032, up from 0% in 2022. Additionally, America is projected to capture over one-quarter (28%) of total global capital expenditures (capex) from 2024-2032, ranking second only to Taiwan (31%). In the absence of the CHIPS Act, the U.S. would have captured only 9% of global capex by 2032, according to the report.

While the report finds investments from the industry—facilitated by CHIPS incentives—are on track to reinvigorate semiconductor manufacturing in America and reinforce U.S. chip supply chains, it also identifies policy actions that will further strengthen supply chains, support R&D and chip design, grow the semiconductor workforce, and ensure CHIPS delivers maximum benefits to America’s economic and national security.

The report also analyzes the efforts underway in other countries to incentivize chip production and innovation and the criticality of ensuring chip companies have open access to global customers and suppliers, among other topics.

“Effective policies, such as the CHIPS and Science Act, are spurring more investments in the U.S. semiconductor industry. These investments will help America grow its share of global semiconductor production and innovation, furthering economic growth and technological competitiveness,” said Rich Templeton, Chairman of the Board at Texas Instruments and SIA board chair. “Continued and expanded government-industry collaboration will help ensure we build on this momentum and continue our next steps forward.”

Other key report findings:

• America’s world-leading 203% projected increase in fab capacity from 2022 to 2032 stands in stark contrast to its modest 11% increase from the previous decade (2012-2022), which ranked last among all major chip-producing regions, according to the SIA/BCG report.
• 
  
• The U.S. share of the world’s chip manufacturing capacity will increase from 10% in 2022—when the CHIPS and Science Act was enacted—to 14% by 2032, marking the first time in decades the U.S. has grown its domestic chip manufacturing footprint relative to the rest of the world. In the absence of CHIPS enactment, the U.S. share would have slipped further to 8% by 2032, according to the report.
• 
  
• The U.S. continues to lead the world in its overall contribution to the global value chain, with strong leadership positions in high value-added areas of semiconductor technology, including chip design, electronic design automation (EDA), and semiconductor manufacturing equipment.

The report also finds industrial policies have the potential to create additional bottlenecks that increase supply chain risk. Certain segments of the semiconductor supply chain are at risk if incentive programs and large-scale industrial policies lead to non-market-based investment, which can result in overconcentration or oversupply. Government incentives should focus on enabling targeted, distributed, market-based investments.

Further, the study highlights the ways in which governments and companies are taking concerted action to increase resilience. The U.S. CHIPS Act committed $39 billion in incentives for semiconductor manufacturing, plus a separate advanced manufacturing investment tax credit. The European Union unveiled the European CHIPS Act, China initiated the third phase of its Integrated Circuit (IC) Industry Investment Fund, and various other incentive programs have emerged in Taiwan, Korea, Japan, India, and around the world. In parallel, companies have made significant investments, in both established and new regions. The report projects around $2.3 trillion in capex in 2024-2032, compared to $720 billion in the decade prior to enactment of the CHIPS Act (2013-2022).

Despite the progress made to strengthen U.S.-based semiconductor manufacturing, additional government policy actions are needed to help ensure America stays on track to address lingering supply chain vulnerabilities and grow its share of fabrication capacity, while also increasing its strength in areas such as advanced logic, design, EDA, and equipment in the face of growing global competition.

“The CHIPS and Science Act has put America on course to significantly strengthen domestic semiconductor production and R&D, but more work is needed to finish the job,” said John Neuffer, SIA president and CEO. “We look forward to working with government leaders to advance policies that broaden the STEM talent pipeline, invest in scientific research, promote free trade and access to global markets, and expand and extend critical CHIPS incentives.”

The CHIPS Act’s manufacturing incentives have sparked substantial announced investments in the U.S. In fact, companies in the semiconductor ecosystem have announced more than 80 new projects across 25 U.S. states—totaling nearly $450 billion in private investments—since the CHIPS Act was introduced. These announced projects will create more than 56,000 jobs in the semiconductor ecosystem and support hundreds of thousands of additional U.S. jobs throughout the U.S. economy.

America Projected to Triple Semiconductor Manufacturing Capacity by 2032, the Largest Rate of Growth in the World

U.S. forecast to increase its domestic chip manufacturing by 203% in the decade following enactment of the CHIPS and Science Act—and expand its share of the w

www.semiconductors.org

 

[coldsolder215]

Who's gonna soak up all this capacity that the market doesn't need? No one knows. Why is my boss telling me American semiconductor professionals are too expensive to employ en masse? Cross that bridge when we get there. The important thing is that these made up numbers prove that we're all doing a great job.

 

[blueone]

Why is my boss telling me American semiconductor professionals are too expensive to employ en masse?

Let's see, whom do I believe? Your boss or Scotten Jones? I think I'll put all of my chips on Scotten. (Sorry for the pun. I just can't seem to resist lately.)

TSMC Arizona Fab Cost Revisited - Semiwiki

Back in May of 2020 I published some comparisons of the cost to run a TSMC fab in Arizona versus their fabs in Taiwan. I found the fab operating cost based on the country-to-country difference to only be 3.4% higher in the US and then I found an additional 3.8% because of the smaller fab…

semiwiki.com

 

[Daniel Nenni]

From what I can remember through my 40 years, semiconductor cost is key. If what TSMC says is true, that it costs more make wafers in the US, that is going to be a problem when we have over supply. Remember, China has gone down to 5nm without EUV. China is also buying up semiconductor equipment with record setting CAPEX and ramping up manufacturing with impunity.

The questions is: Will we have a semiconductor manufacturing capacity oversupply? And if so when?

My guesses would be yes and starting in 2030. The financial types say the semiconductor industry will hit $1T in 2030. Unfortunately ASPs are growing much faster than chip unit sales. An oversupply will certainly fix that, absolutely.

 

[blueone]

From what I can remember through my 40 years, semiconductor cost is key. If what TSMC says is true, that it costs more make wafers in the US, that is going to be a problem when we have over supply.

I have no doubt that it costs more to make wafers in the US than in Taiwan or China. The question is, as a factor of the total cost of a wafer, how big is that additional cost? I don't know. I also don't know how much of a premium high volume customers will pay to have diversified supply chains, if any. I think the evidence is they will pay a non-trivial premium. For example, Apple is probably spending more to ramp iPhone assembly in India than they were with the status quo in China. Will they pay more for chips? Well, if Apple will pay more for assembly as a hedge against geopolitical and supply concentration risks, why would chips be an exception?

Remember, China has gone down to 5nm without EUV.

Is this really a good example? I thought the Chinese yields for non-EUV are in question, and would anyone else tolerate those costs and risks? Or am I wrong?

Anyway, given ASML's production limitations, it looks like there's a valid argument that EUV implementation will be limiter for 2nm and below deployment anyway.

China is also buying up semiconductor equipment with record setting CAPEX and ramping up manufacturing with impunity.

Yes, and people unfamiliar with semiconductors often forget that design innovation can make up for lack of fab process advantages. You can still build one hell of a supercomputer with ~20nm chips, if cost and power targets are relaxed.

The question is: Will we have a semiconductor manufacturing capacity oversupply? And if so when?

My guesses would be yes and starting in 2030. The financial types say the semiconductor industry will hit $1T in 2030. Unfortunately ASPs are growing much faster than chip unit sales. An oversupply will certainly fix that, absolutely.

My guess, which is obviously less expert than yours, is - how could we not have a glut? A lot of these fabs are just never completed? That'll look stupid to investors, so I'd bet on a glut.

 

[TBiggs]

Let see if TSMC can learn and drive and make significant improvements to get to Scotten’s fantasy estimate of cost differential

 

[nghanayem]

Let see if TSMC can learn and drive and make significant improvements to get to Scotten’s fantasy estimate of cost differential

What is fantastical about it? The price of the tools don't double if the final destination is in the US vs anywhere else. All else being equal, sure the shell will be far more expensive, but I think that it could be not as bad as you'd think. As an example a fab in TW will need way more support pillars in the subfab to deal with worse seismic activity reducing the number of tools that can fit into a fab of similar size and raising the cost of building the shell. Given, TSMC F21 seems to be using a copy of the F18 mods, I'm guessing they have an overbuilt. Maybe for phase 3 and beyond they will use a modified or otherwise completely different shell design that drops unnecessary costs? They shouldn't even need more workers either as they will be working the same hours 12hr shifts 2 days on 2 days off that they work in TW. Chems are commodity goods, and power is more reliable in AZ than TW.

The only thing that is crippling is the fact that TSMC AZ depreciation curves will be 5 years behind F18. Since I doubt TSMC wants to sell AZ N5 wafers at fully depreciated fab prices, they will definitely pass down most of those costs. But at the end of the day that is a transient issue. F21 Phase 2 will be in a slightly less bad spot because it will "only" be 4 years behind F18_P3-6.

Considering Micron wants to put 40% of their DRAM production in the US once the Boise R&D/HVM makes Boise into a D1 or Micron JP like site and the very large NY HVM site gets built out. With DRAM being a commodity good Micron would never even entertain the idea of this unless they thought that incentives would even out shell build costs with the deals they get in Singapore/TW/JP and most importantly that sustained operating costs post depreciation would be close enough to not gimp Micron's ability to compete with their government funded Korean competitors that also have large parent corporations.

 

[hskuo]

From what I can remember through my 40 years, semiconductor cost is key. If what TSMC says is true, that it costs more make wafers in the US, that is going to be a problem when we have over supply. Remember, China has gone down to 5nm without EUV. China is also buying up semiconductor equipment with record setting CAPEX and ramping up manufacturing with impunity.

The questions is: Will we have a semiconductor manufacturing capacity oversupply? And if so when?

My guesses would be yes and starting in 2030. The financial types say the semiconductor industry will hit $1T in 2030. Unfortunately ASPs are growing much faster than chip unit sales. An oversupply will certainly fix that, absolutely.

It would not be surprised the wafer cost is higher in US. To improve profit, then they need to let customers pay more for the wafer price to keep the same margin. It could be possible either providing more yielded chips, faster delivery or more wafer outputs, which seem to be very challenging jobs. Or let customers' pay for the geopolitical derisk or "Made in US" ideology.
It rarely happens that overseas fabs delivery higher yield or shorter cycle time than fabs in Taiwan. For example, to cover 10% more cost and keep the same chip cost, 10% more chips need to be produced or 10% higher yielded wafer needed. It is hard to achieve.

 

[Barnsley]

Money Money Money!!!

Plenty for everyone , hopefully those at the top of the pyramid dont get too greedy and squeeze those below too much more.

 

[TBiggs]

Besides cost of the tools, you got the local labor unions. There are stories about how empty the parking lots are at 4pm and the only ones still working are the Taiwanese. Actually one can go see for themselves, park yourself at 303/17 and see the line of trucks leaving and another stream far late at night.

Then there is quality of work and commitment or lack of it of the American workforce.

All this leads to far more issues with schedule and schedule is money!

Word on the street is the TSMC Taiwanese don’tt want to work at the US plant as they have to do twice as much work, their normal work and than train and do the work of their US workers, what fun it must be to work there. Working already is crazy long hours and than you got to do your Americans newcomer job too. The animosity between the US and Taiwanese is so poisoned. Leadership has done nothing to diffuse it either.

 

[Barnsley]

Besides cost of the tools, you got the local labor unions. There are stories about how empty the parking lots are at 4pm and the only ones still working are the Taiwanese. Actually one can go see for themselves, park yourself at 303/17 and see the line of trucks leaving and another stream far late at night.

Then there is quality of work and commitment or lack of it of the American workforce.

All this leads to far more issues with schedule and schedule is money!

Word on the street is the TSMC Taiwanese don’tt want to work at the US plant as they have to do twice as much work, their normal work and than train and do the work of their US workers, what fun it must be to work there. Working already is crazy long hours and than you got to do your Americans newcomer job too. The animosity between the US and Taiwanese is so poisoned. Leadership has done nothing to diffuse it either.

Lots of hate for the American worker!

Are they as bad as you say they are?

Just a question on the East Asian working model that you are glorifying, if it really so great why are the populations in decline , actually same for Western Europe also.

People dont want to be shackled to their job 24/7 to the detriment of their family life , or maybe they more than happy to sacrifice family for work

 

[nghanayem]

Besides cost of the tools, you got the local labor unions. There an ample stories about how empty the parking lots are at 4pm and the only ones still working are the Taiwanese

Fab engineers and techs aren't unionized at any company outside of maybe ST-M which I think recently went on strike or something like that. And since we no longer live in the 1960s they also have round the clock coverage. If union folks are headed home, those are construction workers. Now, maybe those guys work at night in TW, but that doesn't sound right.

Then there is quality of work and commitment or lack of it of the American workforce.

All this leads to far more issues with schedule and schedule is money!

TI in Texas, their Utah conversion, and intel AZ are on schedule. Samsung Texas, and intel Ohio are only running a year late vs TSMC's 2. If others can build fabs in 3 years, intel can do it in 4 years, and TSMC needs 5 years for the first phase and another 5 years for the second phase (with each phase being smaller than a single phase for Taylor or even Ocotillo); that sounds like a TSMC problem to me.

 

[jorgequinonez]

From what I can remember through my 40 years, semiconductor cost is key. If what TSMC says is true, that it costs more make wafers in the US, that is going to be a problem when we have over supply. Remember, China has gone down to 5nm without EUV. China is also buying up semiconductor equipment with record setting CAPEX and ramping up manufacturing with impunity.

The questions is: Will we have a semiconductor manufacturing capacity oversupply? And if so when?

My guesses would be yes and starting in 2030. The financial types say the semiconductor industry will hit $1T in 2030. Unfortunately ASPs are growing much faster than chip unit sales. An oversupply will certainly fix that, absolutely.

All very true about cost. However, if conflict or Cold War goes into high gear with China, having our own supply at home will be critical. The cost then will be the least of our concern.

 

[milesgehm]

Besides cost of the tools, you got the local labor unions. There are stories about how empty the parking lots are at 4pm and the only ones still working are the Taiwanese. Actually one can go see for themselves, park yourself at 303/17 and see the line of trucks leaving and another stream far late at night.

Then there is quality of work and commitment or lack of it of the American workforce.

All this leads to far more issues with schedule and schedule is money!

Word on the street is the TSMC Taiwanese don’tt want to work at the US plant as they have to do twice as much work, their normal work and than train and do the work of their US workers, what fun it must be to work there. Working already is crazy long hours and than you got to do your Americans newcomer job too. The animosity between the US and Taiwanese is so poisoned. Leadership has done nothing to diffuse it either.

I've worked in fabs (actually going into clean rooms and managing process engineers) in CA, TX, MN, Europe, and Singapore). What you are saying is just wrong. American engineers are smart and dedicated.

 

[osnium]

I've worked in fabs (actually going into clean rooms and managing process engineers) in CA, TX, MN, Europe, and Singapore). What you are saying is just wrong. American engineers are smart and dedicated.

Has the quality of American semiconductor process engineers degraded over time in your experience? Intel had process leadership a decade ago and that observation would be consistent with American engineers being better a decade ago than today.

 

[blueone]

Has the quality of American semiconductor process engineers degraded over time in your experience? Intel had process leadership a decade ago and that observation would be consistent with American engineers being better a decade ago than today.

That's not a logical conclusion. Intel was the only US company with fab process leadership, and lost that leadership due to poor senior management and technical leadership decisions, not the quality of its fab engineers.

 

[TBiggs]

American engineers and cultural and expectation is very different than the Far East. Anyone who claims otherwise doesn’t understand the nuances of American versus Korean, Japan, Taiwan, China and Singapore.

As to are they talented and capable, agreed they work different and motivation is different.